Abstract
Introduction: Liver R2* has been shown to be an accurate predictor of liver iron content (LIC) at 1.5 Tesla (T) with a linear relationship between R2* and LIC (Wood JC, Blood 2005). However, imaging at 3T is replacing 1.5T, due to higher signal-to-noise ratio and spatial resolution. Currently, 3T scanners are commonly used in the clinical setting due to its ability to detect more subtle CNS and vascular changes. For patients with sickle cell disease (SCD), who are on cyclic transfusion therapy (CTx) due to cerebral vasculopathy and elevated risk for stroke, the flexibility of 3T scanners generates a need for liver iron measurement at 3T. This study tests the feasibility and accuracy of a novel liver iron quantitation protocol suitable for 3T scanners, in comparison to the standard Ferriscan (R2) at 1.5T.
Methods: The main challenge in measuring liver R2* at 3T is that the relation rates are considerably higher than at 1.5T. In order to maintain the precision of the measurement, a standard gradient echo sequence was developed that uses optimized sampling bandwidths and gradient timings to achieve shorter echo times (minimum echo time of 0.82ms for a 320mm field of view with 8 mm slices and a 96 x 96 acquisition matrix). Patients with SCD, who were on CTx for > 12 month and able to undergo MRI scan without sedation, were consented for the IRB-approved study. On the 3T Siemens Trio scanner, 3 sequences, each consisting of 5 echoes, were run in a single breath-hold (14.5sec) with the initial echo time of the second and third repetitions being offset by one third and two thirds of the inter-echo interval, respectively. Six axial slices through the center of the liver were acquired for each subject. The images were post-processed using in-house software. Briefly, the user outlined the liver in each slice, for each voxel within the defined regions. The software determined which echoes were above the noise level and fitted these to an exponential function. A histogram of the resulting R2* values was then calculated and this was fitted to a Gaussian function to determine the peak of the histogram and hence the mean R2* value. Ferriscan data was obtained using the standard Resonance Health protocol on a 1.5T Siemens Avanto scanner and the data were processed by resonance health.
Results: Seventeen pediatric patients (mean age 12.7, range 8.7-19.2 yrs) had Ferriscan and 3T exams (11 had the exams on the same day, 6 within two months of each other – average delay of 0.9 months). One subject had a liver iron content, as measured by Ferriscan, that was outside of the acceptable range for their calibration (>43 mg/g) and this subject was excluded from analysis. All subjects were able to cooperate with the breath-holding. The typical time to complete the R2* exam was 5 minutes, compared to 15 minutes for Ferriscan. The R2* and Ferriscan showed a relatively strong correlation (R2=0.92), the intercept of 92 Hz is consistent with the published values for normal liver (iron levels < 1mg/g) at 3T (Figure). While more data is required, particularly at intermediate levels of liver iron to validate the calibration, the results imply that this approach provides a quick, robust estimate of liver iron levels on 3T scanners.
Conclusion: Our results indicate that liver iron content can be derived using an optimized gradient echo sequence on a 3T scanner in a clinical setting.
No relevant conflicts of interest to declare.
